Palantir Beats, AI Market Finally Believes.

Mazen Al Sabbagh

— 6 min read
Share

TL;DR

  • Palantir Validates Thesis: $1.94 billion in Q2 revenue, up 93%, with full-year guidance raised to $8.15 billion. The stock registered a 16% premarket gain, breaking a pattern of post-earnings declines.
  • Markets Peak Ahead of Policy: The Dow reached an all-time closing high Monday; Amazon briefly surpassed a $3 trillion valuation. AMD and SpaceX earnings are anticipated tonight.
  • Tao Addresses AI Proofs: Terence Tao, in his first post since July 26, criticized the trend of AI-generated and verified proofs being abandoned without full academic commitment, an oblique reference to recent developments from OpenAI and Astra.
  • White House Framework Ambiguity: Key AI developers—Anthropic, OpenAI, Google, Meta—meet today with the White House to review a model-testing framework, the specifics of which are reportedly still undisclosed to the companies involved.
  • Anthropic's Capital Structure: The Financial Times detailed an intricate financing arrangement, reportedly totaling $200 billion, supporting Anthropic's chip infrastructure through Google, Apollo, Blackstone, and Morgan Stanley.

Lead Story: Palantir Beats, AI Market Finally Believes.

Palantir reported second-quarter revenue of $1.94 billion, a 93% increase year-over-year, exceeding the $1.81 billion consensus. EPS registered at $0.41. The underlying growth suggests a significant uptake of its AI-driven platforms, particularly within the commercial sector.

The core of the market's response was driven by the raised full-year guidance, now projected at $8.15–$8.16 billion—an 82% increase. US commercial revenue guidance surged to $764 million, up 149%, with a Q3 outlook of approximately $2.16 billion, above the $2 billion street estimate. These figures provided critical validation for Palantir's long-term revenue trajectory.

After closing Monday at $125.65, the stock experienced a robust post-market rally, climbing 16% to nearly $146 premarket. This performance marks a notable departure from the previous four quarters, each of which saw the stock decline despite earnings beats. CEO Alex Karp encapsulated the sentiment, stating: "for the first time people believe us."

This strong AI sector validation occurred against a backdrop of broader market milestones. Monday's close saw the Dow Jones Industrial Average reach an all-time high of 53,178.41, up 1.32%. The S&P 500 closed at 7,601.41 (+1.49%) and the Nasdaq at 25,913.90 (+2.13%). Amazon also briefly crossed the $3 trillion intraday valuation mark, becoming the fifth company to do so, though it finished just under at $284.02.

Meanwhile, the KOSPI in Seoul staged a significant rebound, recovering from intraday lows to close up 1.62% at 6,358.95, with the KOSDAQ jumping 5.88%. Samsung and SK Hynix, key components of the global AI chip supply chain, clawed back 3% intraday losses. Goldman Sachs set an aggressive 12,000 KOSPI target, citing depressed valuations, as new leveraged-ETF caps take effect on Wednesday.

As Tuesday's trading commences, futures indicate further gains driven by Palantir and Caterpillar. AMD is up over 2% pre-market ahead of its earnings report tonight, while Brent crude oil prices ticked back towards $84 as Iran clarified its Hormuz Strait talks are proceeding through Oman, not direct engagement with Washington.

In Other News

Terence Tao Breaks Silence on AI Proofs. In a two-part post published early Tuesday, Tao outlined five essential stages for mathematical proof validation. He then articulated a concern over the "disturbing" trend of proofs generated and verified by AI tools that are subsequently abandoned at an intermediate stage, suggesting that those claiming credit for such proofs should commit to their full publication. While neither OpenAI nor its recent ten-proofs release were explicitly named, the description strongly aligns with recent AI-driven mathematical advancements. Scott Aaronson, whose work touches on quantum complexity, remains notably silent on the matter.

White House AI Framework Meeting Faces Transparency Concerns. The Office of the National Cyber Director is hosting representatives from Anthropic, Google, Meta, and OpenAI today to discuss a proposed model-testing framework. However, Semafor reports that these firms have not yet received the final text of the framework, and the White House has declined to confirm whether the document will be publicly released. Benchmarks are reportedly classified—"Just because things are unclassified that doesn't mean we are going to broadcast them to everyone". The Federal Register and NIST websites showed no updates this morning. Reuters noted the tests will assess models' hacking capabilities, following Anthropic's three-incident disclosure and OpenAI's agent escape into Hugging Face, which has since prompted a letter from state attorneys general.

The FT Maps Google's Extensive Financial Engineering for Anthropic. The Financial Times revealed a complex, roughly $200 billion financial structure centered on Anthropic. Over $150 billion of this interlinked capital is reportedly allocated to TPU chips for Anthropic's operations, involving key players like Broadcom, Apollo, Blackstone, and Morgan Stanley. This includes a Morgan Stanley private-credit vehicle designed to acquire chips and lease them to Anthropic, drawing parallels to aircraft financing models. Alphabet's direct investment is projected at $10 billion, with potential to increase to $40 billion based on performance targets. This intricate architecture surfaces as Anthropic's October IPO window approaches.

OpenAI Counter-Files Against Apple's Trade Secrets Allegations. In an August 3 response characterizing Apple's lawsuit as "careless, aggressive, and oddly personal," OpenAI released emails and chat logs. These communications suggest Apple's outside counsel may have mistakenly contacted the wrong individual, confusing general counsel Che Chang with a former Apple employee. OpenAI maintains that a claimed discussion with its GC never occurred.

X / Social Pulse

Reflecting persistent concerns regarding AI's impact on labor markets, Hacker News' dominant organic discussions centered on themes of human displacement and the nuances of AI interaction. "Don't be a meat proxy" garnered 1,748 points, followed by "LLMs reward expertise" at 1,035, and "Prevent cognitive debt by manually retyping LLM-generated code" with 502 points.

The thread discussing OpenAI's "ten-advances" in mathematics reached 557 points, but accumulated 848 comments—a notable divergence suggesting a contentious and deeply debated topic within the community.

Follow-up reporting on the OpenAI super-PAC story traced "The Wire by Acutus," which featured a fictitious reporter "Michael Chen" and had 69% of its articles flagged as AI-generated by Pangram, back to talking points aligning with the $125 million Leading The Future PAC.

The "Pacing the Frontier" letter, advocating for responsible AI development, saw its signatory count increase to 1,346, with John Schulman now among the prominently listed names.

A historical note circulating on the front page: Ray Bradbury's short story "There Will Come Soft Rains" is precisely set on today's date—August 4, 2026.

One to Watch

Key Reports Due: AMD and SpaceX. AMD is scheduled to report earnings tonight, with consensus estimates around $11.3 billion in revenue (+47%) and EPS of $1.61. Options pricing indicates an anticipated ~9% stock movement. Wedbush recently raised its target for AMD to $600 from $450, citing the deployment of next-generation Instinct GPUs in Azure and a projected supply of up to 2GW of MI450 chips for Anthropic starting in 2027. SpaceX is also expected to release its report post-market close. Additionally, any readout from today's White House AI framework meeting could land at any hour, potentially triggering immediate market or policy reactions.

Quick Hits

Palantir's strong quarter provided the tangible revenue validation the AI trade has demanded, propelling markets to new records even before the trading day concluded. Concurrently, the crucial regulatory framework for these powerful AI models undergoes review today by the very companies it impacts, paradoxically before they have even seen its final text. This juxtaposition highlights a persistent asymmetry between rapid market and technological advancement and the slower, more opaque evolution of governance.

Sources

Lead / markets: CNBC on Palantir · Palantir release · Fortune on Karp · Benzinga premarket · Bloomingbit Monday closes · Bloomberg on Amazon $3T · Sedaily KOSPI close · Yahoo Tuesday live blog · FXEmpire on oil · Seeking Alpha AMD preview · Barchart on Wedbush

Policy: Politico via Yahoo on the meeting · Reuters via US News · Political Wire on Semafor · The Next Web on classification · Crowell on SB 947 · CSAC calendar

Models / labs: Tao part 1 · Tao part 2 · OpenAI ten advances · Shtetl-Optimized · FT on Google-Anthropic financing · CNBC on Anthropic IPO · 9to5Mac on OpenAI-Apple · Forbes on OpenAI-Apple · Qwen on Hugging Face · Musk on Grok 4.6 · Valar Atomics

Pulse: HN: meat proxy · HN: LLMs reward expertise · HN: retyping LLM code · HN: ten advances · HN: super PAC · HN: Qwen3.8-Max · HN: Soft Rains · Pacing the Frontier

Lock in. M. mazen@thorterminal.com

Read more